CONTEMPORARY ADVANCES IN PHYSICS 357 



according to (17); the approximation being closer, the higher the 

 temperature and the lower the field.* 



Now there is a very large class of paramagnetic substances of which 

 the susceptibilities at low fieldstrengths conform, over wide ranges of 

 temperature, to equations like (17); and what renders the theory 

 important for our present purposes is, that the ferromagnetic metals 

 at high temperatures enter into that class. To make the test for any 

 substance it is best to plot l/o- as a function of absolute (or Centigrade) 

 temperature. On doing this for nickel beyond the Curie-point (near 

 360° C.) one finds a curve which at first is somewhat bent, but beyond 

 410° passes into a beautiful straight line which continues undeflected 

 to 900°. This line is shown in Fig. 14, together with data for iron 

 beyond its Curie-point at 775°; among these, the points for tempera- 

 tures between 920° and 1395° lie along a straight line which is sharply 

 broken off at each end of that interval, being followed beyond 1395° 

 by what seems to be the beginnings of an entirely different line, 

 and preceded before 920° by a series of points which are well fitted 

 by a pair of straight lines connected with each other at 828°. The 

 data for cobalt beyond its Curie-point at 1130° likewise conform to a 

 pair of connecting straight lines. 



For each of these straight lines one may compute the values of the 

 constants called C and 0; and from these, if one accepts the theory, 

 the values of the moment M of the elementary magnets and the 

 coefficient n of the postulated extra force. In calculating M it is 

 necessary to make an assumption about the number of elementary 

 magnets per unit volume of the metal; assuming that there are as 

 many such as there are atoms, and expressing M in Weiss magnetons, 

 Weiss obtained the values 20.9, 17.4, 28.2 and 7.05 for the four straight 

 lines of iron (in order of increasing temperature); 15.9 aiid 14.55 for 

 those of cobalt; 8 for the solitary straight line of nickel. All these 

 are of the orders of magnitude customarily found in dealing with 

 paramagnetic gases and salts and solutions. The corresponding values 

 of e are 1047, 1063, - 1340, 1543; 1404, 1422; and 645. The 

 corresponding values of n (which is the quotient of 6 by C) are of 

 the order of several thousands. The postulated extra field must 

 therefore be supposed enormou^sly greater than the field //, and even 

 the induction is quite insignificant by comparison with it. In one 

 of these cases (and in many others among the paramagnetic salts, 



* I should state that formulae of the same type as (17) may be derived without 

 assuming that there is a molecular field, provided that we suppose that the distri- 

 bution-in-energy of the atoms in thermal equilibrium is governed not by the equi- 

 partition-law, but by a quantum-law involving a zero point energy. 



